Machine learning techniques enable computers to perform astonishing feats ranging from facial recognition to uncannily accurate predictions about stock market performance.
Big data is part of what has allowed the enormous advancements in artificial intelligence in recent years, and Matt Malloy, newly hired as assistant adjunct professor of electrical and computer engineering, once worked behind the scenes to develop some important and widely used tools for parsing meaningful information from massive internet datasets.
One of those tools helped enable the sometimes-uncanny accuracy of targeted ads—the phenomenon when a product you view from your phone’s e-commerce app then appears in promotions on other websites or in banners advertising goods that seem tailor-made for your tastes. Malloy and colleagues at the analytics company comScore described the dataset in the 2016 paper, “Internet Device Graphs” and more recently in 2019 in a paper titled “Graphing Crumbling Cookies.”
“My work in industry involved developing cutting edge machine learning techniques and implementing them at internet scale,” says Malloy. “We worked on both the tools to collect and gather information, and the algorithms that make inference on those massive datasets.”
Malloy brings a unique blend of industry and academic experience to his new role in the ECE department.
“I’ve wanted to come back to academia for quite some time,” he says.
After receiving his undergraduate degree at UW-Madison in 2004, he completed a master of science degree in electrical engineering at Stanford University before working at the multinational telecommunications company Motorola for three years. Malloy returned to UW-Madison for his PhD (which he finished in 2013) and then spent one year as a postdoctoral scholar. For the past five years, he’s been working in industry as Director and Principle Data Scientist at Comscore, where he led a team of data scientists and software engineers.
In his new position with ECE, Malloy will shift his focus away from research and development and to education as a full-time teacher. He’s leading a full docket of electrical and computer engineering courses, as well as taking over as program director for the department’s popular new accelerated master’s degree program in signal processing and machine learning.
Malloy cut his teeth on teaching during his time as a post doc. He’s always been eager to implement modern techniques to optimize student learning; the course videos he created as the instructor of the ECE 203 class, Signals, Information, and Computation, have been viewed more than 31,000 times since he first recorded the material in 2014.
And while Malloy eagerly embraces educational innovation, and is working to incorporate flipped and blended approaches into many of his courses, his teaching philosophy is grounded in keeping students engaged and excited about the material. He does that by presenting clear objectives along with concrete examples to illustrate the course concepts at work in the real world.
Keeping students engaged is sometimes challenging, especially when covering the abstract and challenging statistical concepts that underpin modern machine learning techniques. But the challenge is worth it when engineers leave UW-Madison with the skills they need to succeed.
“Engineering students need to develop a keen mathematical sense,” says Malloy. “Not just in academia but in industry, as well. Engineers need to be able to break down and solve any problems they encounter.”
Author: Sam Million-Weaver